pubs.acs.org/Langmuir © 2009 American Chemical Society
)
Unanticipated CdC Bonds in Covalent Monolayers on Silicon Revealed by NEXAFS Michael V. Lee,*,† Jonathan R. I. Lee,‡ Daniel E. Brehmer,§ Matthew R. Linford, and Trevor M. Willey‡ ICYS-MANA, National Institute for Materials Science, Japan, ‡Condensed Matter and Materials Division, Lawrence Livermore National Laboratory, Livermore, California, §Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, Menlo Park, California, and Department of Chemistry and Biochemistry, Brigham Young University, Provo, Utah )
†
Received October 9, 2009. Revised Manuscript Received November 19, 2009 Interfaces are crucial to material properties. In the case of covalent organic monolayers on silicon, molecular structure at the interface controls the self-assembly of the monolayers, which in turn influences the optical properties and electrical transport. These properties intrinsically affect their application in biology, tribology, optics, and electronics. We use near-edge X-ray absorption fine structure spectroscopy to show that the most basic covalent monolayers formed from 1-alkenes on silicon retain a double bond in one-fifth to two-fifths of the resultant molecules. Unsaturation in the predominantly saturated monolayers will perturb the regular order and affect the dependent properties. The presence of unsaturation in monolayers produced by two different methods also prompts the re-evaluation of other radical-based mechanisms for forming covalent monolayers on silicon.
Covalent organic monolayers on silicon, first reported in 1993,1 have proven to be useful in a wide range of applications that include biological sensors,2-4 MEM coatings,5 electrochemical detectors,6 photonics,7-9 and nanoscale electronics.10-12 Surface modification with covalent monolayers on silicon has become an important tool for much of nanotechnology. Thermal hydrosilylation was one of the first methods discovered for forming these covalent monolayers.13 Because of its simplicity (i.e., a hydrogen-terminated silicon wafer is immersed in alkene and heated), thermal hydrosilylation has become one of the standard methods for forming covalent monolayers. Thermal hydrosilylation forms a monolayer through a chain reaction. In the chain reaction, an alkene reacts with a silicon radical on an otherwise hydrogen-terminated surface, leading to a carboncentered radical. The carbon radical can then abstract a hydrogen *Corresponding author. E-mail:
[email protected]. (1) Linford, M. R.; Chidsey, C. E. D. J. Am. Chem. Soc. 1993, 115, 12631–12632. (2) Wojtyk, J. T. C.; Morin, K. A.; Boukherroub, R.; Wayner, D. D. M. Langmuir 2002, 18, 6081–6087. (3) de Smet, L. C. P. M.; Stork, G. A.; Hurenkamp, G. H. F.; Sun, Q.; Topal, H.; Vronen, P. J. E.; Sieval, A. B.; Wright, A.; Visser, G. M.; Zuilhof, H.; Sudholter, E. J. R. J. Am. Chem. Soc. 2003, 125, 13916–13917. (4) Bocking, T.; James, M.; Coster, H. G. L.; Chilcott, T. C.; Barrow, K. D. Langmuir 2004, 20, 9227–9235. (5) Ashurst, W. R.; Yau, C.; Carraro, C.; Lee, C.; Kluth, G. J.; Howe, R. T.; Maboudian, R. Sens. Actuators, A 2001, 91, 239–248. (6) Bateman, J. E.; Eagling, R. D.; Worrall, D. R.; Horrocks, B. R.; Houlton, A. Angew. Chem., Int. Ed. 1998, 37, 2683–2685. (7) Boukherroub, R.; Morin, S.; Wayner, D.; Lockwood, D. Phys. Status Solidi A 2000, 182, 117–121. (8) Boukherroub, R.; Morin, S.; Wayner, D.; Bensebaa, F.; Sproule, G.; Baribeau, J.; Lockwood, D. Chem. Mater. 2001, 13, 2002–2011. (9) Buriak, J. M.; Stewart, M. P.; Geders, T. W.; Allen, M. J.; Choi, H. C.; Smith, J.; Raftery, D.; Canham, L. T. J. Am. Chem. Soc. 1999, 121, 11491–11502. (10) Lee, M. V.; Hoffman, M. T.; Barnett, K.; Geiss, J.; Smentkowski, V. S.; Linford, M. R.; Davis, R. C. J. Nanosci. Nanotechnol. 2006, 6, 1639–1643. (11) Yang, L.; Lua, Y.; Tan, M.; Scherman, O. A.; Grubbs, R. H.; Harb, J. N.; Davis, R. C.; Linford, M. R. Chem. Mater. 2007, 19, 1671–1678. (12) Lee, M. V.; Nelson, K. A.; Hutchins, L.; Becerril, H. A.; Cosby, S. T.; Blood, J. C.; Wheeler, D. R.; Davis, R. C.; Woolley, A. T.; Harb, J. N.; Linford, M. R. Chem. Mater. 2007, 19, 5052–5054. (13) Linford, M. R.; Fenter, P.; Eisenberger, P. M.; Chidsey, C. E. D. J. Am. Chem. Soc. 1995, 117, 3145–3155.
1512 DOI: 10.1021/la9038254
atom from the surface to form the next silicon radical. This propagating radical (produced by other methods) has been observed by STM at various stages of “zipping” the double bond from styrene to the surface.14,15 At higher temperatures, the majority of them unzip as well.14 Another less common, but versatile, method for forming covalent monolayers is chemomechanical functionalization.16-18 For chemomechanical functionalization, a silicon surface is submerged in a reactive molecule and the silicon surface is scratched to expose the underlying silicon. Where the silicon is scratched, the molecule reacts with the silicon radicals to form a covalent monolayer. This has been shown to be a facile method for many common functional groups.18 Until now, the theory has held that the chemomechanical reaction of 1-alkenes (CdC) on silicon produces covalent, unsaturated alkyl monolayers (C-C) similar to those produced by thermal hydrosilylation. Although chemomechanical functionalization on the nanoscale produces a smooth pattern, when large areas are functionalized the resulting surface is rough. Both kinds of monolayers have been analyzed by various methods, including wetting, chemical stability, X-ray photoelectron spectroscopy (XPS), X-ray reflectivity, and MIR-IR.13,19,20 Many of these analysis methods, including ATR-IR/MIR-IR, require a smooth surface, precluding their use in the analysis of chemomechanical functionalization monolayers. Generally, the analysis methods qualitatively measure bulk or physical properties. (14) Lopinski, G. P.; Wayner, D. D. M.; Wolkow, R. A. Nature 2000, 406, 48–51. (15) Eves, B. J.; Sun, Q.; Lopinski, G. P.; Zuilhof, H. J. Am. Chem. Soc. 2004, 126, 14318–14319. (16) Linford, M. R. Producing Coated Particles by Grinding in the Presence of Reactive Species. U.S. Patent 6,132,801, Oct 17, 2000. (17) Niederhauser, T. L.; Lua, Y.; Jiang, G.; Davis, S. D.; Matheson, R.; Hess, D. A.; Mowat, I. A.; Linford, M. R. Angew. Chem., Int. Ed. 2002, 41, 2353–2356. (18) Yang, L.; Lua, Y.; Lee, M. V.; Linford, M. R. Acc. Chem. Res. 2005, 38, 933–942. (19) Niederhauser, T. L.; Jiang, G.; Lua, Y.; Dorff, M. J.; Woolley, A. T.; Asplund, M. C.; Berges, D. A.; Linford, M. R. Langmuir 2001, 17, 5889–5900. (20) Buriak, J. Chem. Rev. 2002, 102, 1271–1308.
Published on Web 11/25/2009
Langmuir 2010, 26(3), 1512–1515
Lee et al.
Letter
Figure 1. Generally accepted mechanisms for chemomechanical functionalization and thermal hydrosilylation predict the loss of one degree of unsaturation during the formation of a covalent monolayer on a silicon surface. However, the NEXAFS spectra of these monolayers clearly show the significant retention of unsaturation in monolayers formed by both methods.
The most common exception, IR, is able to qualitatively identify specific chemical groups in the bound monolayers based on vibrational spectroscopy. Until now, only saturated alkyl chains (C-C) have been observed in covalent monolayers on silicon formed from 1-alkenes (CdC) by thermal hydrosilylation or chemomechanical functionalization. Only with a second functional group or an additional degree of unsaturation do functional features appear. In some cases, another method based on electronic spectroscopy that quantitatively evaluates bonding (e.g., π vs σ or π* vs σ*) in the monolayers could clearly supplement the surface analysis performed to date. Near-edge X-ray absorption fine structure (NEXAFS) spectroscopy powerfully probes molecular electronic structure with synchrotron radiation. Photons excite electrons from the core 1s orbitals into the antibonding molecular orbitals, effectively mapping the partial density of unoccupied states. These spectral maps include π* peaks, which are characteristic of carbon sp2 or sp bonds, as well as σ*, characteristic of sp3. For mechanistic investigations, this spectroscopy technique provides a natural complement to vibrational spectroscopy such as IR. NEXAFS has been used in several instances to analyze covalent monolayers on silicon, but these reports have focused on electron transport through the monolayers21 or on bonding to silicon22 but have, under the respective experimental circumstances, correctly dismissed observed π* features to be due to damage from irradiation or spectrum-processing issues. We carefully eliminated both of these possible problems from our measurements and found that a significant C 1s to π* transition peak exists in the spectra for monolayers produced by both the chemomechanical and even the thermal hydrosilylation methods (Figure 1). Preparation of covalent organic monolayers with the same reagent, 1-octene, by thermal hydrosilylation on both rough and smooth Si(100) surfaces as well as by chemomechanical functionalization allows a comparison of the monolayers based on the surface morphology and formation method. The monolayers on rough surfaces were hydrophobic, and XPS analysis produced spectra expected from covalent alkyl monolayers on silicon. Simple checks by wetting and ellipsometry on planar samples (21) Seitz, O.; Vilan, A.; Cohen, H.; Hwang, J.; Haeming, M.; Schoell, A.; Umbach, E.; Kahn, A.; Cahen, D. Adv. Funct. Mater. 2008, 18, 2102–2113. (22) Hu, Y. F.; Boukherroub, R.; Sham, T. K. J. Electron Spectrosc. Relat. Phenom. 2004, 135, 143–147.
Langmuir 2010, 26(3), 1512–1515
Figure 2. NEXAFS spectra of covalent monolayers on silicon formed from reaction with 1-octene by chemomechanical functionalization (which inherently forms a rough surface), thermal hydrosilylation on roughened silicon, and thermal hydrosilylation on a smooth Si(100) surface. A spectrum from gaseous 1-octene is included for reference. The overall step height is proportional to the total amount of carbon in the sample. The peaks at 285 eV come from the promotion of an electron from the ground-state carbon 1s orbital to the π* unsaturated states.
(the sample on Si(100) in Figure 2 had an advancing contact angle of 88° and a thickness of 1 nm by ellipsometry) as well as XPS (Supporting Information) showed them to be typical monolayers.10,18,23 The NEXAFS data were acquired at beamlines 8.2 (monolayers) and 10.1 (gas phase) at the Stanford Synchrotron Radiation Lightsource, and the monolayer spectra were processed in the following manner. The incident flux was adjusted to minimize exposure while obtaining a sufficient signal-to-noise ratio. We estimate the exposure to be 20 μC/mm2).21 No change in π* intensity was observed during two to three successive NEXAFS acquisitions on the same spot on a sample (Supporting Information). The beam was moved to expose fresh surface for each spectrum acquired for sample characterization. The raw monolayer spectra were carefully normalized to their respective scans from the upstream gold grid and subsequently to those from UHV flash-cleaned silicon. The scans from flash-cleaned Si surfaces were acquired directly preceding and/or following each monolayer NEXAFS acquisition. Comparison of the total electron yield (TEY) of monolayers versus clean Si clearly shows π* character in the monolayers (Supporting Information). No irregularity that would produce a false peak during normalization was observed in the upstream Au grid scans nor in the clean Si scans. The π* peak is also present in auger yield NEXAFS, a technique that is generally less sensitive to beam normalization (Supporting Information). On the basis of prevailing theories, C 1s NEXAFS spectroscopy was expected to give a flat line in the region near 285 eV, the spectral region indicative of unsaturated carbon (1s f π* transition). Figure 2 presents spectra from gas-phase 1-octene as a reference as well as monolayers prepared from 1-octene by chemomechanical functionalization on silicon, by thermal hydrosilylation on rough silicon, and also by thermal hydrosilylation on Si(100). Instead of a flat line, NEXAFS, which is very sensitive to the difference between saturated and unsaturated carbon, revealed significant π* character (Figure 2, peak ∼285 eV) in monolayers synthesized by both methods. The size of the π* peak relative to the step edge height allows comparative quantification. Comparison to published spectra for condensed ethylene28 suggests that for monolayers prepared by chemomechanical functionalization on rough silicon, thermal hydrosilylation on rough silicon, or thermal hydrosilylation on (25) Sieval, A.; Demirel, A.; Nissink, J.; Linford, M.; van der Maas, J.; de Jeu, W.; Zuilhof, H.; Sudholter, E. Langmuir 1998, 14, 1759–1768. (26) Lin-Vien, D.; Colthup, N. B.; Fateley, W. G.; Grasselli, J. G. The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules; Academic Press: San Diego, CA; 1991; pp 74, 84-85. (27) Scheres, L.; Arafat, A.; Zuilhof, H. Langmuir 2007, 23, 8343–8346. (28) Matsui, F.; Yeom, H. W.; Matsuda, I.; Ohta, T. Phys. Rev. B 2000, 62, 5036. (29) From adsorbed ethylene NEXAFS in ref 27, if there were six more saturated carbons per molecule then the step edge would be 4 times as tall and the height of the 285 eV peak is about 0.8 times this value. The ratios of the ∼285 eV peak maximum to the step edge height in the monolayers in Figure 2 are 0.34, 0.28, and 0.13.
1514 DOI: 10.1021/la9038254
smooth silicon at least one-third, one-fourth, and one-eighth, respectively, of the adsorbed molecules retain a double bond.29 The peaks from the monolayers are obviously broad and complex rather than narrow Gaussian-shaped like those for condensed ethylene, so these numbers will certainly underestimate the content of double bonds. If we integrate the peak areas, we may get a more accurate picture of the number of molecules with a double bond. Comparison of the integrated area of the π* peaks to gaseous 1-octene spectra (Figure 2) produces the respective values of two-fifths, two-fifths, and one-fifth for the same monolayers.30 The doublebond absorbance can be muted for a condensed layer, so this may still underestimate the proportion of unsaturation. Nevertheless, these values put a lower bound on the double bond content and confirm that the mechanism producing CdC bonds in each method is a major pathway for the formation of the monolayers. Direct probing of the binding and structure for monolayers formed from 1-alkenes and silicon has generally relied on IR spectroscopy, which shows the presence of alkyl chains and the absence of any other characteristic peaks.6,13,25 One may wonder why the CdC feature in thermal hydrosilylation monolayers is so strong in NEXAFS spectroscopy yet has not been observed by IR. The SiCdC stretch and the alkyl-substituted RCHdCHR stretch have been reported to be weak.26 Indeed, for IR spectra of different monolayers with a double bond in each bound molecule, Scheres et al. reported that the SiCdC stretch is “rather small and is in fact nearly impossible to detect by IRRAS.” Further spectra by ATR-IR were required to produce a spectrum with a small but clear peak.27 A fractional monolayer such as those represented in Figure 2 would have an even smaller peak that could easily be buried in noise in IRRAS. Another possible contributing factor is that recent characterization by IR has focused on monolayers formed at lower temperatures, initiated by light or chemical means,20 which would likely increase the yield of the saturated product. Better-packing monolayers have been produced by these lower-temperature methods, which could be explained by a lower content of unsaturated molecules.27 For chemomechanical functionalization monolayers, this unsaturated peak provides important insight into the chemisorption. (30) Calculated values were 0.40, 0.38, and 0.21.
Langmuir 2010, 26(3), 1512–1515
Lee et al.
Previous indirect investigations have confirmed a radical-based mechanism31 and covalent binding.19 Now, the direct measurement of chemical structure by NEXAFS reveals unsaturation extant in the monolayers. If unsaturation is retained by the donation of a hydrogen to the surface, then at least 40% of the monolayer is formed in this fashion. If instead disproportionation is the source of the unsaturated molecules, then at least 80% of the monolayer molecules are bound by this process. The unsaturation in Figure 2 represents either a major pathway or the principal pathway for chemomechanical monolayer formation. Even more exciting are the implications for thermal hydrosilylation, which have been analyzed quite extensively but until now unsaturation had not been seen. The double bonds could spring from the collision and simultaneous termination by disproportionation of two propagating radicals. However, that explanation would require the initiation of significantly more radicals for the formation of monolayers than is generally accepted; the magnitude of the unsaturation suggests an alternate explanation. Another explanation that would not require a greater degree of initiation would be an alternate pathway for the propagation of the radical. One H from the surface could combine with one from the carbon-centered radical intermediate to evolve H2 and leave a CdC bond (Scheme 1). A basic analysis of bond strengths (i.e., breaking two C-H bonds and one C-C bond to form one CdC and one H-H bond suggests that the enthalpy of the unsaturated product is only ∼62 kJ/mol higher than that of the reversible saturated product (Supporting Information).32 Because of the loss of hydrogen, this portion of the propagation would not be reversible and would be entropically favored, which would enable it to compete with the lower-enthalpy reversible reaction. (31) Jiang, G.; Niederhauser, T. L.; Fleming, S. A.; Asplund, M. C.; Linford, M. R. Langmuir 2004, 20, 1772–1774. (32) Lide, D. R., Ed. CRC Handbook of Chemistry and Physics, 85th ed.; CRC: New York; 2004.
Langmuir 2010, 26(3), 1512–1515
Letter
This is consistent with observations made by Lopinski et al. in their STM observations. At low temperatures, the propagating radical formed lines of styrene on Si(100); when the surfaces were heated, most of the lines desorbed but some were stable even at elevated temperatures.14 Lower temperatures would produce relatively fewer unsaturated molecules, but the lines that included one would not desorb. NEXAFS shows significant sp2 character in monolayers that are formed from 1-alkenes chemisorbed on silicon via thermal hydrosilylation or chemomechanical functionalization. In contrast to generally accepted mechanisms, at least one-fifth or two-fifths of the molecules in layers formed by these methods retain a double bond. The high level of unsaturation in thermal hydrosilylation requires a pathway that simultaneously continues radical propagation and produces double bonds. These results represent a major departure from our previous understanding of these most basic of covalent monolayers. These results also illustrate the usefulness of NEXAFS spectroscopy as a complement to IR. Acknowledgment. M.V.L. acknowledges support by the World Premier International Research Center (WPI) Initiative on Materials Nanoarchitectonics, MEXT, Japan. J.R.I.L. and T.M.W. acknowledge funding by the Office of Basic Energy Sciences (OBES), Materials Sciences, U.S. DOE. This work was partially performed at the Lawrence Livermore National Laboratory under contract DE-AC52-07NA27344. The Stanford Synchrotron Radiation Lightsource is a national user facility operated by SLAC National Accelerator Laboratory on behalf of the U.S. DOE, OBES. Supporting Information Available: Extended experimental details; additional checks for normalization, calibration, and beam damage; angle-resolved NEXAFS; and NEXAFS for a homologous series of alkenes. This material is available free of charge via the Internet at http://pubs.acs.org.
DOI: 10.1021/la9038254
1515
